private void handleExternalTables(
final HiveMetaStoreBridge dgiBridge,
final HiveEventContext event,
final LinkedHashMap<Type, Referenceable> tables)
throws HiveException, MalformedURLException {
List<Referenceable> entities = new ArrayList<>();
final Entity hiveEntity = getEntityByType(event.getOutputs(), Type.TABLE);
Table hiveTable = hiveEntity.getTable();
// Refresh to get the correct location
hiveTable = dgiBridge.hiveClient.getTable(hiveTable.getDbName(), hiveTable.getTableName());
final String location = lower(hiveTable.getDataLocation().toString());
if (hiveTable != null && TableType.EXTERNAL_TABLE.equals(hiveTable.getTableType())) {
LOG.info("Registering external table process {} ", event.getQueryStr());
final ReadEntity dfsEntity = new ReadEntity();
dfsEntity.setTyp(Type.DFS_DIR);
dfsEntity.setName(location);
SortedMap<Entity, Referenceable> inputs =
new TreeMap<Entity, Referenceable>(entityComparator) {
{
put(dfsEntity, dgiBridge.fillHDFSDataSet(location));
}
};
SortedMap<Entity, Referenceable> outputs =
new TreeMap<Entity, Referenceable>(entityComparator) {
{
put(hiveEntity, tables.get(Type.TABLE));
}
};
Referenceable processReferenceable =
getProcessReferenceable(dgiBridge, event, inputs, outputs);
String tableQualifiedName =
dgiBridge.getTableQualifiedName(dgiBridge.getClusterName(), hiveTable);
if (isCreateOp(event)) {
processReferenceable.set(AtlasClient.REFERENCEABLE_ATTRIBUTE_NAME, tableQualifiedName);
}
entities.addAll(tables.values());
entities.add(processReferenceable);
event.addMessage(new HookNotification.EntityUpdateRequest(event.getUser(), entities));
}
}
private LinkedHashMap<Type, Referenceable> createOrUpdateEntities(
HiveMetaStoreBridge dgiBridge,
HiveEventContext event,
Entity entity,
boolean skipTempTables,
Table existTable)
throws Exception {
Database db = null;
Table table = null;
Partition partition = null;
LinkedHashMap<Type, Referenceable> result = new LinkedHashMap<>();
List<Referenceable> entities = new ArrayList<>();
switch (entity.getType()) {
case DATABASE:
db = entity.getDatabase();
break;
case TABLE:
table = entity.getTable();
db = dgiBridge.hiveClient.getDatabase(table.getDbName());
break;
case PARTITION:
partition = entity.getPartition();
table = partition.getTable();
db = dgiBridge.hiveClient.getDatabase(table.getDbName());
break;
}
db = dgiBridge.hiveClient.getDatabase(db.getName());
Referenceable dbEntity = dgiBridge.createDBInstance(db);
entities.add(dbEntity);
result.put(Type.DATABASE, dbEntity);
Referenceable tableEntity = null;
if (table != null) {
if (existTable != null) {
table = existTable;
} else {
table = dgiBridge.hiveClient.getTable(table.getDbName(), table.getTableName());
}
// If its an external table, even though the temp table skip flag is on,
// we create the table since we need the HDFS path to temp table lineage.
if (skipTempTables
&& table.isTemporary()
&& !TableType.EXTERNAL_TABLE.equals(table.getTableType())) {
LOG.debug(
"Skipping temporary table registration {} since it is not an external table {} ",
table.getTableName(),
table.getTableType().name());
} else {
tableEntity = dgiBridge.createTableInstance(dbEntity, table);
entities.add(tableEntity);
result.put(Type.TABLE, tableEntity);
}
}
event.addMessage(new HookNotification.EntityUpdateRequest(event.getUser(), entities));
return result;
}
private boolean isExternal(Table table) {
return TableType.EXTERNAL_TABLE.equals(table.getTableType());
}
@Override
public final CatalogProtos.TableDescProto getTable(String databaseName, final String tableName)
throws CatalogException {
org.apache.hadoop.hive.ql.metadata.Table table = null;
HiveCatalogStoreClientPool.HiveCatalogStoreClient client = null;
Path path = null;
String storeType = null;
org.apache.tajo.catalog.Schema schema = null;
KeyValueSet options = null;
TableStats stats = null;
PartitionMethodDesc partitions = null;
//////////////////////////////////
// set tajo table schema.
//////////////////////////////////
try {
// get hive table schema
try {
client = clientPool.getClient();
table = HiveCatalogUtil.getTable(client.getHiveClient(), databaseName, tableName);
path = table.getPath();
} catch (NoSuchObjectException nsoe) {
throw new UndefinedTableException(tableName);
} catch (Exception e) {
throw new TajoInternalError(e);
}
// convert HiveCatalogStore field schema into tajo field schema.
schema = new org.apache.tajo.catalog.Schema();
List<FieldSchema> fieldSchemaList = table.getCols();
boolean isPartitionKey = false;
for (FieldSchema eachField : fieldSchemaList) {
isPartitionKey = false;
if (table.getPartitionKeys() != null) {
for (FieldSchema partitionKey : table.getPartitionKeys()) {
if (partitionKey.getName().equals(eachField.getName())) {
isPartitionKey = true;
}
}
}
if (!isPartitionKey) {
String fieldName =
databaseName
+ CatalogConstants.IDENTIFIER_DELIMITER
+ tableName
+ CatalogConstants.IDENTIFIER_DELIMITER
+ eachField.getName();
TajoDataTypes.Type dataType =
HiveCatalogUtil.getTajoFieldType(eachField.getType().toString());
schema.addColumn(fieldName, dataType);
}
}
// validate field schema.
HiveCatalogUtil.validateSchema(table);
stats = new TableStats();
options = new KeyValueSet();
options.putAll(table.getParameters());
options.remove("EXTERNAL");
Properties properties = table.getMetadata();
if (properties != null) {
// set field delimiter
String fieldDelimiter = "", nullFormat = "";
if (properties.getProperty(serdeConstants.FIELD_DELIM) != null) {
fieldDelimiter = properties.getProperty(serdeConstants.FIELD_DELIM);
} else {
// if hive table used default row format delimiter, Properties doesn't have it.
// So, Tajo must set as follows:
fieldDelimiter = "\u0001";
}
// set null format
if (properties.getProperty(serdeConstants.SERIALIZATION_NULL_FORMAT) != null) {
nullFormat = properties.getProperty(serdeConstants.SERIALIZATION_NULL_FORMAT);
} else {
nullFormat = "\\N";
}
options.remove(serdeConstants.SERIALIZATION_NULL_FORMAT);
// set file output format
String fileOutputformat =
properties.getProperty(hive_metastoreConstants.FILE_OUTPUT_FORMAT);
storeType = HiveCatalogUtil.getStoreType(fileOutputformat);
if (storeType.equalsIgnoreCase("TEXT")) {
options.set(
StorageConstants.TEXT_DELIMITER, StringEscapeUtils.escapeJava(fieldDelimiter));
options.set(StorageConstants.TEXT_NULL, StringEscapeUtils.escapeJava(nullFormat));
} else if (storeType.equals("RCFILE")) {
options.set(StorageConstants.RCFILE_NULL, StringEscapeUtils.escapeJava(nullFormat));
String serde = properties.getProperty(serdeConstants.SERIALIZATION_LIB);
if (LazyBinaryColumnarSerDe.class.getName().equals(serde)) {
options.set(StorageConstants.RCFILE_SERDE, StorageConstants.DEFAULT_BINARY_SERDE);
} else if (ColumnarSerDe.class.getName().equals(serde)) {
options.set(StorageConstants.RCFILE_SERDE, StorageConstants.DEFAULT_TEXT_SERDE);
}
} else if (storeType.equals("SEQUENCEFILE")) {
options.set(
StorageConstants.SEQUENCEFILE_DELIMITER,
StringEscapeUtils.escapeJava(fieldDelimiter));
options.set(StorageConstants.SEQUENCEFILE_NULL, StringEscapeUtils.escapeJava(nullFormat));
String serde = properties.getProperty(serdeConstants.SERIALIZATION_LIB);
if (LazyBinarySerDe.class.getName().equals(serde)) {
options.set(StorageConstants.SEQUENCEFILE_SERDE, StorageConstants.DEFAULT_BINARY_SERDE);
} else if (LazySimpleSerDe.class.getName().equals(serde)) {
options.set(StorageConstants.SEQUENCEFILE_SERDE, StorageConstants.DEFAULT_TEXT_SERDE);
}
}
// set data size
long totalSize = 0;
if (properties.getProperty("totalSize") != null) {
totalSize = Long.parseLong(properties.getProperty("totalSize"));
} else {
try {
FileSystem fs = path.getFileSystem(conf);
if (fs.exists(path)) {
totalSize = fs.getContentSummary(path).getLength();
}
} catch (IOException ioe) {
throw new TajoInternalError(ioe);
}
}
stats.setNumBytes(totalSize);
}
// set partition keys
List<FieldSchema> partitionKeys = table.getPartitionKeys();
if (null != partitionKeys) {
org.apache.tajo.catalog.Schema expressionSchema = new org.apache.tajo.catalog.Schema();
StringBuilder sb = new StringBuilder();
if (partitionKeys.size() > 0) {
for (int i = 0; i < partitionKeys.size(); i++) {
FieldSchema fieldSchema = partitionKeys.get(i);
TajoDataTypes.Type dataType =
HiveCatalogUtil.getTajoFieldType(fieldSchema.getType().toString());
String fieldName =
databaseName
+ CatalogConstants.IDENTIFIER_DELIMITER
+ tableName
+ CatalogConstants.IDENTIFIER_DELIMITER
+ fieldSchema.getName();
expressionSchema.addColumn(new Column(fieldName, dataType));
if (i > 0) {
sb.append(",");
}
sb.append(fieldSchema.getName());
}
partitions =
new PartitionMethodDesc(
databaseName, tableName, PartitionType.COLUMN, sb.toString(), expressionSchema);
}
}
} finally {
if (client != null) client.release();
}
TableMeta meta = new TableMeta(storeType, options);
TableDesc tableDesc = new TableDesc(databaseName + "." + tableName, schema, meta, path.toUri());
if (table.getTableType().equals(TableType.EXTERNAL_TABLE)) {
tableDesc.setExternal(true);
}
if (stats != null) {
tableDesc.setStats(stats);
}
if (partitions != null) {
tableDesc.setPartitionMethod(partitions);
}
return tableDesc.getProto();
}
private boolean isManaged(Table table) {
return TableType.MANAGED_TABLE.equals(table.getTableType());
}